Title:
BIOMETRIC MATCHING ENGINE
Kind Code:
A1


Abstract:
The present disclosure concerns a method of identifying a biometric record of an individual in a database (108), the database comprising at least first and second sets of records, each set comprising at least one record, the method comprising: receiving by a processing device (102) at least first and second input biometric samples of said individual; performing on the records of said first set a first matching process comprising a first filtering operation followed by a second filtering operation, and performing on the records of said second set a second matching process comprising said second filtering operation followed by said first filtering operation, wherein said first filtering operation comprises comparing said first input biometric sample to a first reference biometric sample of each record, and said second filtering operation comprises comparing said second input biometric sample to a second reference biometric sample of each record; and identifying a biometric record of said individual based on results of the first and second matching processes.



Inventors:
Partington, Alastair Ross (Hatfield, GB)
Astrom, Per Anders Matteus (Antibes, FR)
Bataller, Cyrille (Mougins, FR)
Application Number:
14/478772
Publication Date:
06/04/2015
Filing Date:
09/05/2014
Assignee:
ACCENTURE GLOBAL SERVICES LIMITED
Primary Class:
International Classes:
G06F17/30; G06K9/00
View Patent Images:
Related US Applications:



Foreign References:
WO2006135394A22006-12-21
Primary Examiner:
BUI, THUY T
Attorney, Agent or Firm:
Harrity & Harrity, LLP (11350 Random hills road Suite 600 Fairfax VA 22030)
Claims:
1. 1-15. (canceled)

16. A method comprising: receiving, by a processing device, a first input biometric sample and a second input biometric sample; determining, by the processing device, a first characteristic associated with the first input biometric sample and a second characteristic associated with the second input biometric sample; grouping, by the processing device and after receiving the first input biometric sample and the second input biometric sample, a plurality of records to form a first set of records and a second set of records, the grouping being based on the first characteristic and the second characteristic; performing, by the processing device and on the first set of records, a first matching process based on a first threshold; performing, by the processing device and on the second set of records, a second matching process based on a second threshold; and identifying, by the processing device, a record, of the plurality of records, based on a result of the first matching process and a result of the second matching process.

17. The method of claim 16, where the first input biometric sample comprises a first type of biometric sample, and where the second input biometric sample comprises a second type of biometric sample that is different from the first type of biometric sample.

18. The method of claim 17, where the first type of input biometric sample comprises one of: an image of a face, a fingerprint, an iris scan, an image of a signature, an image of a finger vein, an image of a palm vein, or a voice sample.

19. The method of claim 16, where grouping the plurality of records includes: comparing a parameter relating to at least one reference biometric sample, associated with each record of the plurality of records, with at least one threshold, and grouping the plurality of records to form the first set of records and the second set of records based on the comparing.

20. The method of claim 19, where the parameter comprises a quality measurement parameter associated with the reference biometric sample associated with each record of the plurality of records.

21. The method of claim 20, further comprising: determining the quality measurement parameter, associated with the reference biometric sample associated with each record of the plurality of records, based on one or more of: an age of an individual associated with the first input biometric sample, a gender of the individual, a size of an image corresponding to the reference biometric sample, a sharpness of the image, a viewing angle of the image, or a contrast in the image.

22. The method of claim 16, further comprising: determining an order for performing the first matching process and the second matching process, where the first matching process and the second matching process are performed in the determined order.

23. A device comprising: one or more processors to: receive a first input biometric sample and a second input biometric sample; determine a first characteristic associated with the first input biometric sample and a second characteristic associated with the second input biometric sample; group, after receiving the first input biometric sample and the second input biometric sample, a plurality of records to form a first group of records and a second group of records, the grouping being based on the first characteristic and the second characteristic; perform a first matching process on the first group of records based on a first threshold; perform a second matching process on the second group of records based on a second threshold; and identify a record, of the plurality of records, based on a result of the first matching process and a result of the second matching process.

24. The device of claim 23, where, when performing the second matching process, the one or more processors are to: perform the second matching process in parallel with the first matching process.

25. The device of claim 23, where the first input biometric sample comprises a first type of biometric sample; and where, when grouping the plurality of records, the one or more processors are to: determine that a first record, of the plurality of records, is associated with a reference biometric sample that comprises the first type of biometric sample, and include the first record in the first group of records based on the first record being associated with the reference biometric sample that comprises the first type of biometric sample.

26. The device of claim 25, where, when grouping the plurality of records, the one or more processors are further to: determine that a second record, of the plurality of records, is not associated with a reference biometric sample that comprises the first type of biometric sample, and include the second record in the second group of records based on the second record not being associated with the reference biometric sample that comprises the first type of biometric sample.

27. The device of claim 23, where each record, of the plurality of records, is associated with a respective reference biometric sample; where the one or more processors are further to: determine a quality score for each record based on one or more characteristics associated with the respective reference biometric sample associated with each record; and where, when grouping the plurality of records, the one or more processors are to: group the plurality of records to form the first group of records and the second group of records based on the quality score determined for each record.

28. The device of claim 23, where each record, included in the first group of records, is associated with a particular type of reference biometric sample; where the one or more processors are further to: assign the first matching process to the first group of records based on each record, included in the first group of records, being associated with the particular type of reference biometric sample; and where, when performing the first matching process, the one or more processors are to: perform the first matching process on the first group of records based on the first matching process being assigned to the first group of records.

29. The device of claim 23, where, when performing the first matching process, the one or more processors are to: perform a plurality of filtering operations on the first group of records.

30. A non-transitory computer-readable medium storing instructions, the instructions comprising: one or more instructions that, when executed by one or more processors, cause the one or more processors to: receive a first input biometric sample and a second input biometric sample, the first input biometric sample and the second input biometric sample being associated with an individual to be identified; determine a first characteristic associated with the first input biometric sample and a second characteristic associated with the second input biometric sample; partition, after receiving the first input biometric sample and the second input biometric sample, a plurality of records into a first group of records and a second group of records, the partitioning being based on the first characteristic and the second characteristic; perform a first matching process on the first group of records based on a first threshold; perform a second matching process on the second group of records based on a second threshold; and determine the identity of the individual based on a result of the first matching process and a result of the second matching process.

31. The non-transitory computer-readable medium of claim 30, where the instructions further comprise: one or more instructions that, when executed by the one or more processors, cause the one or more processors to: determine that the first matching process and the second matching process are to be performed at least partially in series; and determine an order for performing the first matching process and the second matching process, where the first matching process and the second matching process are performed based on the determined order.

32. The non-transitory computer-readable medium of claim 30, where the first input biometric sample comprises a first image associated with a first viewing angle; where a record, of the plurality of records, is associated with a second image associated with a second viewing angle; and where the one or more instructions to group the plurality of records include: one or more instructions that, when executed by the one or more processors, cause the one or more processors to: selectively include the record in the first group of records when a difference between the first viewing angle and the second viewing angle is within a particular range of values, the record being included in the second group of records when the difference is not within the particular range of values.

33. The non-transitory computer-readable medium of claim 30, where the first input biometric sample comprises a first type of biometric sample and the second input biometric sample comprises a second type of biometric sample; and where the one or more instructions to group the plurality of records include: one or more instructions that, when executed by the one or more processors, cause the one or more processors to: include a record, of the plurality of records, in the first group of records when the record is associated with a reference biometric sample that comprises the first type of biometric sample, and include the record in the second group of records when the record is associated with a reference biometric sample that comprises the second type of biometric sample.

34. The non-transitory computer-readable medium of claim 30, where the first type of input biometric sample comprises one of: an image of a face of the individual, a fingerprint of the individual, an iris scan of the individual, an image of a signature of the individual, an image of a finger vein of the individual, an image of a palm vein of the individual, or a voice sample of the individual.

35. The non-transitory computer-readable medium of claim 30, where each record, of the plurality of records, is associated with a respective reference biometric sample, the instructions further comprising: one or more instructions that, when executed by the one or more processors, cause the one or more processors to: determine a quality score for each record based on one or more characteristics associated with the respective reference biometric sample associated with each record; and where the one or more instructions to group the plurality of records include: one or more instructions that, when executed by the one or more processors, cause the one or more processors to: group the plurality of records to form the first group of records and the second group of records based on the quality score determined for each record.

Description:

The present application is related to co-pending U.S. patent applications entitled BIOMETRIC TRAINING AND MATCHING ENGINE having Attorney Docket No. 33836.98.0023, and to BIOMETRIC MATCHING SYSTEM having Attorney Docket No. 33836.98.0024, both of which are filed on even date herewith.

FIELD

The present disclosure relates to identifying a record in a biometric database based on one or more input biometric samples.

BACKGROUND

The use of biometric data for the identification of individuals is increasingly becoming the preferred choice in many environments due to the relative difficulty in fraudulently replicating the data. For example, due to increasing fraud involving payment cards such as credit cards, it has been proposed to use biometric data, such as for example fingerprints, to identify customers in shops or supermarkets to allow a payment transaction to be initiated. As a further example, biometric data is increasing used for identifying individuals authorized to enter restricted areas, such as gyms, border controls or vehicles. Furthermore, criminal databases have long been used for identifying individuals based on biometric data, such as a fingerprint or facial image taken at a crime scene.

To identify individuals, a biometric sample is obtained and compared to the records of a database, until a match is found. In the majority of applications, speed is of the essence. For example, if a user is at the checkout of a supermarket, or at a border control, an identification delay of more than several seconds may be considered unacceptable. A further requirement is that there are very few errors, i.e. very few false positive and false negative results. Indeed, if a customer at the checkout of a supermarket can not be identified, or is wrongly identified, this could lead to the customer being unable to make the payment, or to the wrong person being billed.

However, there is a technical problem in increasing the speed of identification and/or in reducing the error rate in current biometric identification systems.

SUMMARY

It is an aim of embodiments of the present disclosure to at least partially address one or more problems in the prior art.

According to one aspect, there is provided a method of identifying a biometric record of an individual in a database, the database comprising at least first and second sets of records, each set comprising at least one record, the method comprising: receiving by a processing device at least first and second input biometric samples of said individual; performing on the records of said first set a first matching process comprising a first filtering operation followed by a second filtering operation, and performing on the records of said second set a second matching process comprising said second filtering operation followed by said first filtering operation, wherein said first filtering operation comprises comparing said first input biometric sample to a first reference biometric sample of each record, and said second filtering operation comprises comparing said second input biometric sample to a second reference biometric sample of each record; and identifying a biometric record of said individual based on results of the first and second matching processes.

According to one embodiment, said first input biometric sample and said first reference biometric sample of each record are of a first type and said second input biometric sample and said second reference biometric sample of each record are of a second type different to said first type.

According to another embodiment, said first type and said second type are each either: a photo of the face of the individual; or a fingerprint of the individual; or an iris scan of the individual; or an image of a signature of the individual; or a finger vein or palm vein image of the individual; or a voice sample of the individual.

According to another embodiment, said first and second filtering operations each comprise eliminating one or more records based on the corresponding comparison.

According to another embodiment, the method further comprises, prior to performing said first and second matching processes, partitioning said database into a first partition comprising said first set of records and a second partition comprising said second set of records by comparing a parameter relating to at least one reference biometric sample of each record of said database with at least one threshold.

According to another embodiment, said partitioning is further based on a parameter relating to either or both of said first and second input biometric samples.

According to another embodiment, said parameter is a quality measurement of said reference biometric sample of each record of said database.

According to another embodiment, said partitioning is based on a quality measurement of at least one of the first and second input biometric samples and on a quality measurement of a reference biometric sample of each record.

According to another embodiment, the method further comprises determining said quality measurement based on either: the age of the individual of the biometric sample; or the gender of the individual of the biometric sample; or the size of an image of the biometric sample; the sharpness of an image of the biometric sample; or the viewing angle of an image of the biometric sample; or the contrast in an image of the biometric sample; or any combination of the above.

According to another embodiment, said first matching process comprises at least one filtering threshold determined based on metadata extracted from at least one record of said first set, and said second matching process comprises at least one filtering threshold determined based on metadata extracted from at least one record of said second set.

According to a further aspect, there is provided a computer readable medium storing a computer program that, when executed by a processor, causes the above method to be implemented.

According to a further aspect, there is provided a biometric matching device comprising: a database comprising first and second sets of records, each set comprising at least one record; and a processor configured to: perform on the records of said first set a first matching process comprising a first filtering operation followed by a second filtering operation, and perform on the records of said second set a second matching process comprising said second filtering operation followed by said first filtering operation, wherein said first filtering operation comprises comparing said first input biometric sample to a first reference biometric sample of each record, and said second filtering operation comprises comparing said second input biometric sample to a second reference biometric sample of each record; and identify a biometric record of said individual based on results of the first and second matching processes.

According to one embodiment, said processing device is further configured to partition said database into a first partition comprising said first set of records and a second partition comprising said second set of records by comparing a parameter relating to at least one reference biometric sample of each record of said database with at least one threshold.

According to a further aspect, there is provided a biometric matching system comprising: one or more biometric capturing devices for capturing at least one input biometric sample of an individual; and the above biometric matching device.

According to a further aspect, there is provided a method of identifying a biometric record of an individual in a database of biometric records, the method comprising: in response to the reception by a processing device of an input biometric sample of said individual, partitioning said database by said processing device into at least first and second partitions each comprising at least one of said biometric records, said partitioning comprising comparing a parameter relating to at least one reference biometric sample of each record of said database with at least one threshold value; evaluating by said processing device according to a first matching process whether the input biometric sample matches a reference biometric sample from each record of said first partition; evaluating by said processing device according to a second matching process whether the input biometric sample matches a reference biometric sample from each record of said second partition; and identifying the biometric record of said individual based on results of the first and second matching processes.

The details of various embodiments are set forth in the accompanying drawings and the description below. Other potential features will become apparent from the description, the drawings and the claims.

BRIEF DESCRIPTION OF THE DRAWINGS

The foregoing and other purposes, features and aspects of the disclosure will become apparent from the following detailed description of example embodiments, given by way of illustration and not limitation with reference to the accompanying drawings, in which:

FIG. 1 schematically illustrates a biometric identification system according to an example embodiment;

FIG. 2 schematically illustrates an electronic device according to an example embodiment;

FIG. 3 illustrates a portion of a biometric database according to an example embodiment;

FIG. 4 is a flow diagram showing operations in a method of identifying a biometric record according to an example embodiment;

FIG. 5 is a flow diagram showing operations in a method of a biometric matching process according to an example embodiment;

FIG. 6 is a flow diagram showing operations for assigning matching processes to records of a database according to an example embodiment;

FIGS. 7A and 7B illustrate examples of extracted metadata according to an example embodiment;

FIG. 8 is a diagram schematically representing an example of three matching processes applied to partitions of a biometric database;

FIG. 9 illustrates an example of a memory; and

FIG. 10 is a flow diagram illustrating a sequence of operations involving the addition of a new database partition.

Throughout the figures, like features have been labelled with like reference numerals.

DETAILED DESCRIPTION

FIG. 1 illustrates a biometric identification system 100 according to an example embodiment.

System 100 comprises a matching engine 102, which receives a biometric input sample SBin from a capturing device (D1) 104. The matching engine 102 may also receive a further biometric input sample S′Bin from a further capturing device (D2) 106. The capturing device 104 is for example a visible light or infra-red camera, a fingerprint sensor, microphone or other detector suitable for capturing a biometric sample of an individual. Input biometric sample SBin could for example be a photo of the face, a fingerprint, an iris scan, an image of a signature, a finger vein or palm vein image, a voice sample, or any other form of biometric data of an individual. Input biometric sample S′Bin is for example also one of this list, but a different one to the sample SBin, such that two types of biometric samples can be used to evaluate a match. Additional capturing devices (not illustrated in FIG. 1) may be provided for capturing further biometric samples of the individual.

In some cases the individual is present at the detector 104 and/or 106 and submits the biometric input sample, for example by presenting their face to a camera or placing a finger on a fingerprint detector. In other cases, the biometric data may be retrieved from another source, such as from the scene of a crime in the case of a latent fingerprint, or from a surveillance video image.

The matching engine 102 receives the one or more biometric input samples from detector 104 and/or 106. In particular, the samples SBin and S′Bin may be digital values transmitted to matching engine 102 via a data connection. The matching engine 102 could be located next to the detectors 104, 106, in which case the data connection could be a USB (universal serial bus) connection, Bluetooth connection, or similar wired connection. Alternatively, the matching engine 102 could be located remotely from the detectors 104, 106, and the data connection could be formed of a wired and/or wireless connection comprising a local area network (LAN), a metropolitan data network (MAN), wide area network (WAN), and/or the internet.

The matching engine 102 has access to a database (DB) 108 storing biometric records each associated with one or more reference biometric samples. The matching engine 102 searches the database for a record having a reference biometric sample matching the input biometric sample or samples. A result (R) is provided on an output 110 of the matching engine 102, and for example simply indicates whether or not a match was found. In alternative embodiments, the result R could contain data associated with the matching record, such as a reference number of the matching record, the identity, such as name, of the individual associated with the matching record, or other data.

The matching engine 102 is for example used for identifying a customer at supermarket checkout or a passport holder at a border control, performing criminal suspect identification, or for performing secure premises access control, such as at the entry to a gym or airport lounge.

FIG. 2 illustrates an apparatus 200 that is suitable for implementing the matching engine 102 of FIG. 1.

Apparatus 200 for example comprises a processing device 202, which is in communication with an instruction memory 204, a memory device 206 having a portion 208 for storing biometric samples, and in some embodiments a communications interface 210, a display 212 and one or more input devices 214. The processing device 202 is also in communication with the database 108 of FIG. 1 that stores the records containing reference biometric samples. The processing device 202 may comprises a microprocessor, microcontroller, digital signal processor, or appropriate combinations thereof, and executes instructions stored in the instruction memory 204, which could be a volatile memory such as DRAM (dynamic random access memory), or another type of memory. In some embodiments, the processing device 202 may comprise a plurality of processors that operate in parallel.

The memory device 206 is for example a non-volatile memory, such as a hard disk drive or FLASH drive. The communications interface 210 for example provides a connection to the capturing devices 104, 106, and may be a USB interface, or network interface, for example providing a wired or wireless connection to a communications network such as the internet. The display 212, as well as one or more input devices such as a keyboard or mouse, may be provided for allowing an administrator to control the operation of the matching engine 102, to download software updates, etc.

FIG. 3 illustrates a portion of the database 108 of FIG. 1, storing biometric records according to an example embodiment, each record corresponding to a different individual and containing a number of reference biometric samples of the record holder of The record holders of the records stored in the database will depend on the particular application, but could correspond to members of a gym, employees of an office, or convicted criminals.

In FIG. 3, three biometric records are shown as an example, having references “0001”, “0002” and “0003” respectively indicated in a field 302. Of course in practise the database is likely to contain hundreds or thousands of records. Each biometric record is associated with a corresponding record holder, but for security reasons, the database 108 for example only identifies these individuals by a reference number. A separate table, for example stored by the matching engine 102, may indicate the mapping between the reference numbers of field 302 and personal details of the corresponding record holder, such as name, address, account details etc., depending on the application.

A field 304 for example comprises a digital image of the face of the record holder, a field 306 for example comprises a digital image of the fingerprint of the record holder, a field 308 for example comprises a digital image of an iris scan of the record holder, and a field 310 for example comprises a digital image of the signature of the record holder. Of course, in alternative examples of the biometric database, only some of these fields may be present and/or addition fields comprising other biometric data could be included.

In the example of FIG. 3, not all records comprise a sample in each field 304 to 310. For example, some of the record holders may not have provided some of the reference samples. In particular, only records 0001 and 0002 comprise images of the face of the record holder in field 304, labelled “image 1A” and “image 2A” respectively. Furthermore, only records 0002 and 0003 comprise fingerprint images in field 306, labelled “image2B” and “image3B” respectively, and only records 0001 and 0003 comprise iris scan images in field 308, labelled “image1C” and “image3C” respectively. All three records comprise signature images in field 310, labelled “image1D”, “image2D” and “image3D” respectively.

The one or more input biometric samples to be compared with the reference biometric samples of the database 108 comprise at least a digital representation of the biometric data corresponding to one or more of the fields contained in the matching record of the database. Indeed, an input photo of the face of the individual could not be used to identify the record 0003 given that this record does not contain a corresponding reference image.

FIG. 4 is a flow diagram showing operations in a method of matching biometric data samples according to an example embodiment. These operations are, for example, implemented by the matching engine 102 of FIG. 1.

In a first operation 402, at least one input biometric sample is received by the matching engine, for example from the capturing device 104 and/or 106 of FIG. 1. For example, these samples are stored in the memory portion 208 of FIG. 2.

In a subsequent operation 404, one of at least two different matching processes is assigned to each record of the database. Each matching process for example defines the particular routines that are used to compare the biometric samples, the order in which the types of biometric samples are processed and/or the thresholds used to keep or eliminate match candidates. As will be described in more detail below, the database comprises a plurality of sets of records. For example, based on pre-defined rules, each record of the database is assigned a particular matching process, those records that are assigned the first matching process forming the first set of records, and those records that are assigned the second matching process forming the second set of records. Alternatively, the database is partitioned into sets that group together records having similar properties, as will be explained in more detail below, and a matching process is assigned to each partition. All the records in a partition are assigned the same matching process, which is different for each partition. Some more detailed examples of the various rules that may be used for assigning matching processes to each record and/or for partitioning the database are provided below.

In subsequent operations 406A and 406B, the matching processes are executed on the records of each set of records. In particular, a first matching process is performed on the first set of records of the database in operation 406A, and a second matching process is performed on a second set of records of the database, as shown in operation 406B. One or more further matching processes may also be performed on further database partitions, as indicated by dashed arrow 406C. The operations 406A, 406B and the one or more further operations 406C may be performed in parallel as represented in FIG. 4, in which case the matching engine 102 of FIG. 1 for example comprises multiple processors, one for implementing each matching process. Alternatively, these operations may be performed in series, or in any order. One of skill in the art will appreciate that while only two database partitions are discussed with respect to FIG. 4, any number of database partitions may exist to store corresponding records without departing from the scope of this disclosure.

After operations 406A, 406B and optionally 406C, the results of these matching processes are provided in an operation 408, indicating whether or not a matching record was found. In particular, one or more matching operations applied during each matching process will result in corresponding similarity scores indicating the similarity between the input biometric sample and the reference biometric sample of each record. These scores are for example used to determine whether or not each record is a match. Generally either no matching record or a single matching record will be found for each request based on one or more input biometric samples. If more than one matching record is identified, this may indicate duplication in the database or an error, and thus such an event is for example signalled to a moderator.

FIG. 5 is a flow diagram showing examples of operations in a biometric matching process applied in operation 406A or 406B of FIG. 4, which is for example performed by the matching engine 102 of FIG. 1.

In an operation 502, the matching engine 102 loads one or more input biometric samples and some or all of the reference biometric data samples of a corresponding partition ready for comparison. For example, the samples are loaded into a memory of the matching engine 102.

In a subsequent operation 504, a first filtering operation is used to filter the records of the corresponding set of records based on a comparison of the input biometric sample with a corresponding sample of each record, to filter out at least some of the records. The first filtering operation is for example chosen to filter out around 50 percent of the records very quickly and with relatively few false negatives. The filter is for example associated with a threshold that determines the records to be filtered in or filtered out, based on a similarity score determined for the record. A permissive threshold, for example relatively low, can be chosen such that a relatively high number of records are filtered in. A restrictive threshold, for example relatively high, can be chosen such that a relatively high number of records are filtered out. Generally, the threshold is chosen such that, while there may be a high likelihood that a non-matching record is filtered in, there is a very low risk that the matching record is erroneously filtered out during this operation. As an example, assuming that during the first filtering operation a similarity score of the input biometric sample with each record is provided on a scale of 1 to 100, it may be determined that any record scoring less than 50 can be filtered out as it is very unlikely that such a low-scoring record corresponds to a match. Of course, it may be quite likely that a lot of non-matching records score over such a threshold, but the aim of the first filter is for example to reduce the number or records as much as possible.

In a subsequent operation 506, a second filtering operation is used to analyse the remaining records to identify, if present, the record with a matching biometric sample. The comparison performed in this operation may or may not be based on the same input biometric sample as used in the first filtering operation 504. The second filtering operation is for example chosen to have very low false negative and false positive results, and is thus for example slower to execute per record than the first filtering operation. For example, this second filtering operation returns just one or a few of the best matching records. In one example, a similarity score determined in the second filtering operation is provided on a scale of 1 to 100, and a match is only considered if the record reaches a score of at least 90, which is for example known to be very rare in the case of a non-matching record.

In a subsequent operation 508, it is determined whether or not a matching record has been identified. For example, in some embodiments, the second filtering operation 506 only indicates a match if there is relatively high certainty. In alternative embodiments, the second filtering operation 506 always outputs the best matching record, and also indicates the level of certainty that the record is a match, i.e. that the input biometric data sample came from the record holder. In such an embodiment, operation 508 may involve comparing the level of certainty with a threshold in order to judge whether there is a match. For example, a match is considered to have occurred if there is at least 99 percent certainty. Alternatively, the certainty level of the matching record from each database partition could be used to select the best matching record.

After operation 508, if a match was found from the partition being processed, the next operation is 510, in which this result of this matching process is provided. This result is for example compared to the results of the other matching processes applied to the other partitions in order to generate the result R at the output of the matching engine. Alternatively, if no matching record is found, the next operation is 512, in which a no-match is indicated.

The flow diagram of FIG. 5 provides just one example of a type of matching process that could be applied in order to find a record in the database 108 having one or more reference biometric samples matching one or more input biometric samples. There are various features of this process that can be modified to provide alternative matching processes, such as the particular filtering operations 504 and 506, the thresholds applied in the operation 504, and the type of biometric data sample compared in operations 504, 506, which could be the same or different. Furthermore, rather than the two-phase approach involving the first and second filtering operations, alternative matching processes could involve applying only the second filtering operation, or an additional filtering operations after the second filtering operation, the final filtering operation for example reducing the number of records to just one, or indicating the best matches.

The particular techniques used to compare the biometric samples and detect a match will be known to those skilled in the art, and are for example based on cascaded tests. For example, fingerprint and face recognition is discussed in the publication “Intelligent Biometric Techniques in Fingerprint and Face Recognition”, Jain, L. C. et al. and “Partially Parallel Architecture for AdaBoost-Based Detection With Haar-like Features”, Hiromote et al., the contents of which are hereby incorporated by reference to the extent allowable by the law.

FIG. 6 is a flow diagram illustrating operations for assigning a matching process to each record of the database according to an example embodiment.

In a first operation 600, one or more input samples is received.

In an operation 602, metadata corresponding to the reference biometric samples is extracted from each database record. For example, the metadata comprises a quality measurement of the sample, as will be described in more detail below.

In a subsequent operation 604, metadata is optionally extracted from the input biometric sample or samples. The extracted metadata is for example of the same type as the data extracted from the database records in operation 602.

In a subsequent operation 606, the database is optionally partitioned based on the metadata, by classifying each record of the database into one of a number of partitions. In particular, each time one or more new input biometric samples is received, this for example triggers a new partitioning of the database. In other words, for each search operation, the partitions, and in particular the records that each partition contains, are dynamically re-defined. In some embodiments, the database is partitioned only based on metadata extracted from the database records, in which case the database partitioning may be performed before the input biometric sample is received. Alternatively, the database is partitioned based on a combination of the metadata extracted from the database records and from the metadata extracted from the one or more input biometric samples. The number of database partitions is for example defined by rules set by a system operator, and will depend on factors such as the types of biometric samples and the available matching processes. The number of partitions is for example at least two. There could be several hundreds or even thousands of partitions, there being no upper limit. In some cases the number of partitions may be determined automatically by the matching engine. Partitioning the database is explained in more detail below with reference to FIGS. 7A and 7B.

In a subsequent operation 608, a matching process is assigned to each database partition, or to each record if the database has not been partitioned, based on the metadata. The matching processes are for example assigned based on rules set by the system operator, as will be described in more detail below. In some cases, each record could be assigned a personalized matching process based on the extracted metadata. For example, the filter threshold of each matching process could be calculated based on the extracted metadata.

A subsequent operation 610 involves distributing tasks. For example, in the case that the matching processes for each database partition/record are to be performed in parallel, this involves instructing the corresponding processors accordingly. Alternatively, if the matching processes are to be performed at least partially in series, task distribution for example involves determining the order that the matching processes are to be executed.

FIGS. 7A and 7B respectively illustrate examples of metadata extracted from database records and from biometric input samples.

With reference first to FIG. 7A, a table 700 shows in rows some examples of metadata that may be extracted from a few of the biometric reference samples of database 300. For example, for the photo of face “image1A” of record “0001”, the metadata for example comprises the age and gender of the record holder, in this example having the respective values 66 and male. This information is for example obtained from the record holder.

Furthermore, data relating to the image may be extracted.

For example the image size may be extracted as the number of pixels in the rectangular image, which in the example of “image1A” is for example 1200 by 1600 pixels.

A sharpness level may be evaluated on a scale of 1 to 10 using techniques known to those skilled in the art, which in the example of “image1A” is equal to 8.

A viewing angle may be determined, a zero angle for example indicating that the face is head-on to the camera, a positive angle indicating a face turned to the right, and a negative angle indicating a face turned to the left. In the example of “image1A”, the angle is for example 5 degrees.

A contrast level, for example on a scale of 1 to 20, may also be evaluated by techniques that will be known to those skilled in the art. In the example of “image1A”, the value is 11.

It will be apparent to those skilled in the art that only some of these values, and/or additional values, could be extracted from the database samples. Furthermore, metadata indicating the particular types of biometric reference samples present in each record is for example extracted.

The records of the database are partitioned based on the extracted metadata. A few examples of such partitioning operations will now be provided.

In one example, the records are classified into partitions and/or assigned a matching process based on the particular reference biometric samples that they contain. For example, all records containing an image of a face are placed in a first partition and/or assigned a first matching process, and all those without are placed in a second partition and/or assigned a second matching process.

Additionally or alternatively, the database is for example partitioned and/or matching processes are assigned to each record based on the quality of the samples, which indicates the facility by which matching may be performed on the data. For example, it may be considered that biometric samples from record holders over 60 are of poorer quality and thus harder to analyse than samples from those under 60. In this case, all records for which the age of the record holder is over this threshold are placed in a one partition and/or assigned a first matching process, and all those under this threshold are placed in another partition and/or assigned another matching process.

Additionally or alternatively, one or more properties of the image, such as image size, sharpness, viewing angle or contrast can be used as quality indicators, and used to partition the database and/or assign matching processes to each record. For example, all records having a sharpness of greater than 7 are placed in one partition and/or assigned a first matching process, and all those with a sharpness of less than 7 are placed in another partition and/or assigned another matching process.

Alternatively, a quality score may be determined for each record, for example on a scale of 0 to 10, indicating an overall quality rating of the biometric sample based on the various parameters available. An example of this score is shown in the right-hand column of table 700. This quality score may then be used to partition the database and/or assign a matching process to each record, for example into good quality samples having scores between 6 and 10, and low quality samples having scores between 0 and 5.

In one particular example, each record of the database is classified into one of four partitions as follows, based on the presence of an iris scan and of a fingerprint, and on the quality of the fingerprint:

    • partition 1: Iris scan+no fingerprint
    • partition 2: Iris scan+good quality fingerprint
    • partition 3: Iris scan+low quality fingerprint
    • partition 4: All remaining records (no iris scan)

Once the database is partitioned, a matching process is for example assigned to each partition using predefined rules, based on the biometric reference samples. For example, the matching processes for the four partitions above are based on the following samples used during the first and second filtering operations of FIG. 5, and in some cases during subsequent third and fourth filtering operations. It is assumed that all records having good quality face images.

    • partition 1: 1st filter: iris; 2nd filter: face
    • partition 2: 1st filter: fingerprint with restrictive threshold; 2nd filter: fingerprint; 3rd filter: iris; 4th filter: face.
    • partition 3: 1st filter: fingerprint with permissive threshold; 2nd filter: Iris; 3rd filter: fingerprints; 4th filter: face.
    • partition 4: 1st filter: fingerprint; 2nd filter: face.

With reference to FIG. 7B, a table 750 shows an example of metadata that could be extracted from two input biometric samples “input1A” and “input1B”, which are for example a face image and a fingerprint respectively. In this example, the extracted metadata is for example the image size, the image sharpness, the viewing angle and an overall quality score, which are determined using the same criteria as described above in relation to FIG. 7A. Some or all of this metadata may be used to partition the database, in addition to the metadata associated with each record.

For example, it may be determined that if the input biometric sample and a reference biometric sample are a face image having a similar viewing angle, a fast filtering operation can be applied to the images with a restrictive threshold, whereas if the viewing angles are different, a slower filtering operation should be used, with a more permissive threshold. Thus the records may be partitioned based on the difference between the viewing angles of the face image.

Alternatively or additionally, the records could be classified into partitions and/or assigned a matching process based on a combination of the quality scores, for example by performing a multiplication of the quality score of the input biometric sample with the quality score of the corresponding records.

As another example, the following four partitions could be used:

    • Partition 1: Δθ<10°; Qi.Qr>50
    • Partition 2: Δθ>10°; Qi.Qr>50
    • Partition 3: Δθ<10°; Qi.Qr<50
    • Partition 4: Δθ>10°; Qi.Qr<50
      where Δθ is the difference between the viewing angles of the input biometric samples and the reference biometric samples, Qi is the quality score of the input biometric reference, and Qr is the quality scores of the reference biometric samples.

The matching process assigned to each of these four partitions may use a first filtering operation having a filtering threshold adapted accordingly. For example, partition 1 may have a restrictive threshold, partitions 2 and 3 average thresholds, and partition 4 a permissive threshold.

FIG. 8 is a diagram very schematically representing an example of three matching processes applied to partitions of a biometric database. A region bounded by an oval 800 represents all of the biometric records of the database, while a partition P1 is represented by an oval 802 with diagonal shading inside oval 800, and a partition P2 is represented by an oval 803 with vertical shading inside oval 800. The remaining region inside oval 800 is labelled 804, and represents all the records not falling within partitions P1 and P2, which thus form a default partition.

Matching processes 806, 808 and 810 are respectively assigned to partitions P1 and P2, and to the default partition 804, and are represented by triangles in FIG. 8. It is assumed that three input biometric samples are received, in the form of a fingerprint, an iris scan, and a face image. Matching process 806 for example applies three filtering operations, starting with a fingerprint filter, followed by an iris filter, and then finishing with a face filter. Each filter allows certain records to be eliminated. The matching process 808 for example applies three filtering operations, starting with a fast fingerprint filter, followed by an accurate and thus slower fingerprint filter, and then finishing with an iris filter. The matching process 810 for example applies two filtering operations, starting with a iris filter and followed by a fingerprint filter.

Thus, in the example of FIG. 8, depending on the partition in which a record falls, the order in which the biometric data types are processed will vary. For example, a record will be either processed by a fingerprint filter followed by an iris filter if it falls within partition P1, or vice verse if it falls within the default partition 804. For example, the records of partition P1 have relatively high quality fingerprint images that are quick to process, whereas the records of the default partition 804 have relatively low quality fingerprints that are slow to process, and thus by changing the filtering order based on the fingerprint quality, the overall processing time can be reduced.

FIG. 9 illustrates an example of a memory 900, for example forming part of the memory device 206 of FIG. 2, storing rules for forming database partitions, and definitions of corresponding matching processes. In particular, a memory location 902 stores the rule corresponding to partition P1, and an associated location 904 stores the matching process MP1 associated with partition P1. Similarly, a memory location 906 stores the rule corresponding to partition P2, and an associated location 908 stores the matching process MP2 associated with partition P2. Depending on the number of partitions, further rules may be defined, and assuming a total of N partitions, memory locations 910, 912 store the rule and matching process MPN respectively for a partition PN. A memory location 914 for example stores the default matching process DMP to be applied to all records not falling into one of the other partitions.

For example, each of the rules defining the partitions P1 to PN defines a distinct subset of the records of the database. Alternatively, to generate the partitions, rules may be applied in a certain order, starting with the most restrictive. For example, assuming that a quality level is used to partition the database, a partition P1 may be generated first to comprise all records with a quality over 8, then a partition P2 may be generated to comprise any of the remaining records having a quality over 6, then a partition P3 may be generated to comprise any of the remaining records having a quality over 4, and any remaining records for example fall in the default partition.

The definition of the partitioning rules in memory 900 has the feature of allowing new records to be added to the biometric database without the need of recreating the partitions at that time. Indeed, the partitioning of the new records will occur dynamically, at the time of a subsequent matching task, based on the existing rules.

Furthermore, the partitioning rules may be adapted dynamically and rapidly, for example in response to new records entered into the database. Indeed, this involves only the adjustment of an existing rule in memory 900, and/or the addition of a new rule within memory 900, to define a new partition. For example, a new set of biometric records added to the database may have particularly high fingerprint qualities, allowing a certain matching strategy to be applied to these records. A new rule could be used to group these records in a new partition based on a relatively high threshold for fingerprint quality.

FIG. 10 is a flow diagram illustrating a sequence of operations performed for example using processing device 202 of FIG. 2, and involving the addition of a new database partition.

In an operation 1002, a matching task is performed in response to the reception of one or more input biometric samples, involving creating partitions P1 to PN based on corresponding rules, and then applying matching processes MP1 to MPN to the partitions P1 to PN, and a default matching process to the remaining records of the database.

In a subsequent operation 1004, a new partition PN+1 is defined, for example, in order to improve performance, by adding a new rule and corresponding matching process MPN+1 in the memory 900 described above.

Then, in a subsequent matching task 1006, in response to the reception of one or more new biometric input samples, the partitions P1 to PN+1 are created based on the corresponding rules, and the matching processes MP1 to MPN are applied to partitions P1 to PN as before, and additionally the matching process MPN+1 is applied to the new partition PN+1. The default matching process is also applied to the remaining records.

Thus, the matching processes applied to certain records of the database may be dynamically altered between matching tasks in a simple fashion by simply defining a new rule and matching process in memory 900. The definition of an existing partition may be updated in a similar manner.

A feature of applying different matching processes to different sets of records of the biometric database is that the processes may be adapted to the samples of sets of records. In particular, the order that the types of biometric data are processed may be adapted such that a relatively large number of records are filtered out quickly, without a high number of false negative decisions. In this way, certain partitions may be processed quickly, and the choice of matching process is not constrained by the worst case records of the database.

A feature of partitioning the database in response to new input biometric samples being received is that the database partitions may be altered and new partitions added in a quick and simple fashion.

While a number of specific embodiments of devices and methods of the present disclosure have been provided above, it will be apparent to those skilled in the art that various modifications and alternatives could be applied.

For example, it will be apparent to those skilled in the art that the examples of matching processes applied to the records of the database partitions are merely a few such examples, and that other matching processes could be used.

Furthermore, it will be apparent to those skilled in the art that other criteria could be used to partition the database of biometric records.

Embodiments of the subject matter and the operations described in this specification can be implemented in digital electronic circuitry, or in computer software, firmware, or hardware, including the structures disclosed in this specification and their structural equivalents, or in combinations of one or more of them. Embodiments of the subject matter described in this specification can be implemented as one or more computer programs, i.e., one or more modules of computer program instructions, encoded on computer storage medium for execution by, or to control the operation of, data processing apparatus. Alternatively or in addition, the program instructions can be encoded on an artificially-generated propagated signal, e.g., a machine-generated electrical, optical, or electromagnetic signal, which is generated to encode information for transmission to suitable receiver apparatus for execution by a data processing apparatus. A computer storage medium can be, or be included in, a computer-readable storage device, a computer-readable storage substrate, a random or serial access memory array or device, or a combination of one or more of them. Moreover, while a computer storage medium is not a propagated signal, a computer storage medium can be a source or destination of computer program instructions encoded in an artificially-generated propagated signal. The computer storage medium can also be, or be included in, one or more separate physical components or media (e.g., multiple CDs, disks, or other storage devices).

The operations described in this specification can be implemented as operations performed by a data processing apparatus on data stored on one or more computer-readable storage devices or received from other sources.

The term “data processing apparatus” encompasses all kinds of apparatus, devices, and machines for processing data, including by way of example a programmable processor, a computer, a system on a chip, or multiple ones, or combinations, of the foregoing The apparatus can include special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit). The apparatus can also include, in addition to hardware, code that creates an execution environment for the computer program in question, e.g., code that constitutes processor firmware, a protocol stack, a database management system, an operating system, a cross-platform runtime environment, a virtual machine, or a combination of one or more of them. The apparatus and execution environment can realize various different computing model infrastructures, such as web services, distributed computing and grid computing infrastructures.

A computer program (also known as a program, software, software application, script, or code) can be written in any form of programming language, including compiled or interpreted languages, declarative or procedural languages, and it can be deployed in any form, including as a stand-alone program or as a module, component, subroutine, object, or other unit suitable for use in a computing environment. A computer program may, but need not, correspond to a file in a file system. A program can be stored in a portion of a file that holds other programs or data (e.g., one or more scripts stored in a markup language document), in a single file dedicated to the program in question, or in multiple coordinated files (e.g., files that store one or more modules, sub-programs, or portions of code). A computer program can be deployed to be executed on one computer or on multiple computers that are located at one site or distributed across multiple sites and interconnected by a communication network.

The processes and logic flows described in this specification can be performed by one or more programmable processors executing one or more computer programs to perform actions by operating on input data and generating output. The processes and logic flows can also be performed by, and apparatus can also be implemented as, special purpose logic circuitry, e.g., an FPGA (field programmable gate array) or an ASIC (application-specific integrated circuit).

Processors suitable for the execution of a computer program include, by way of example, both general and special purpose microprocessors, and any one or more processors of any kind of digital computer. Generally, a processor will receive instructions and data from a read-only memory or a random access memory or both. The essential elements of a computer are a processor for performing actions in accordance with instructions and one or more memory devices for storing instructions and data. Generally, a computer will also include, or be operatively coupled to receive data from or transfer data to, or both, one or more mass storage devices for storing data, e.g., magnetic, magneto-optical disks, or optical disks. However, a computer need not have such devices. Moreover, a computer can be embedded in another device, e.g., a mobile telephone, a personal digital assistant (PDA), a mobile audio or video player, a game console, a Global Positioning System (GPS) receiver, or a portable storage device (e.g., a universal serial bus (USB) flash drive), to name just a few. Devices suitable for storing computer program instructions and data include all forms of non-volatile memory, media and memory devices, including by way of example semiconductor memory devices, e.g., EPROM, EEPROM, and flash memory devices; magnetic disks, e.g., internal hard disks or removable disks; magneto-optical disks; and CD-ROM and DVD-ROM disks. The processor and the memory can be supplemented by, or incorporated in, special purpose logic circuitry.

To provide for interaction with a user, embodiments of the subject matter described in this specification can be implemented on a computer having a display device, e.g., a CRT (cathode ray tube) or LCD (liquid crystal display) monitor, for displaying information to the user and a keyboard and a pointing device, e.g., a mouse or a trackball, by which the user can provide input to the computer. Other kinds of devices can be used to provide for interaction with a user as well; for example, feedback provided to the user can be any form of sensory feedback, e.g., visual feedback, auditory feedback, or tactile feedback; and input from the user can be received in any form, including acoustic, speech, or tactile input. In addition, a computer can interact with a user by sending documents to and receiving documents from a device that is used by the user; for example, by sending web pages to a web browser on a user's client device in response to requests received from the web browser.

Embodiments of the subject matter described in this specification can be implemented in a computing system that includes a back-end component, e.g., as a data server, or that includes a middleware component, e.g., an application server, or that includes a front-end component, e.g., a client computer having a graphical user interface or a Web browser through which a user can interact with an implementation of the subject matter described in this specification, or any combination of one or more such back-end, middleware, or front-end components. The components of the system can be interconnected by any form or medium of digital data communication, e.g., a communication network. Examples of communication networks include a local area network (“LAN”) and a wide area network (“WAN”), an inter-network (e.g., the Internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks).

A system of one or more computers can be configured to perform particular operations or actions by virtue of having software, firmware, hardware, or a combination of them installed on the system that in operation causes or cause the system to perform the actions. One or more computer programs can be configured to perform particular operations or actions by virtue of including instructions that, when executed by data processing apparatus, cause the apparatus to perform the actions.

The computing system can include clients and servers. A client and server are generally remote from each other and typically interact through a communication network. The relationship of client and server arises by virtue of computer programs running on the respective computers and having a client-server relationship to each other. In some embodiments, a server transmits data (e.g., an HTML page) to a client device (e.g., for purposes of displaying data to and receiving user input from a user interacting with the client device). Data generated at the client device (e.g., a result of the user interaction) can be received from the client device at the server.

While this specification contains many specific implementation details, these should not be construed as limitations on the scope of any inventions or of what may be claimed, but rather as descriptions of features specific to particular embodiments of particular inventions. Certain features that are described in this specification in the context of separate embodiments can also be implemented in combination in a single embodiment. Conversely, various features that are described in the context of a single embodiment can also be implemented in multiple embodiments separately or in any suitable sub-combination. Moreover, although features may be described above as acting in certain combinations and even initially claimed as such, one or more features from a claimed combination can in some cases be excised from the combination, and the claimed combination may be directed to a sub-combination or variation of a sub-combination.

Similarly, while operations are depicted in the drawings in a particular order, this should not be understood as requiring that such operations be performed in the particular order shown or in sequential order, or that all illustrated operations be performed, to achieve desirable results. In certain circumstances, multitasking and parallel processing may be advantageous. Moreover, the separation of various system components in the embodiments described above should not be understood as requiring such separation in all embodiments, and it should be understood that the described program components and systems can generally be integrated together in a single software product or packaged into multiple software products.

Thus, particular embodiments of the subject matter have been described. Other embodiments are within the scope of the following claims. In some cases, the actions recited in the claims can be performed in a different order and still achieve desirable results. In addition, the processes depicted in the accompanying figures do not necessarily require the particular order shown, or sequential order, to achieve desirable results. In certain implementations, multitasking and parallel processing may be advantageous.